Exact estimation of biodiesel cetane number (CN) from its fatty acid methyl esters (FAMEs) profile using partial least square (PLS) adapted by artificial neural network (ANN)

Abstract

Cetane number (CN) is among the most important properties of biodiesel because it quantifies combustion speed or in better words, ignition quality. Experimental measurement of biodiesel CN is rather laborious and expensive. However, the high proportionality of biodiesel fatty acid methyl esters (FAMEs) profile with its CN is very appealing to develop straightforward and inexpensive computerized tools for biodiesel CN estimation. Unfortunately, correlating the chemical structure of biodiesel to its CN using conventional statistical and mathematical approaches is very difficult. To solve this issue, partial least square (PLS) adapted by artificial neural network (ANN) was introduced and examined herein as an innovative approach for the exact estimation of biodiesel CN from its FAMEs profile. In the proposed approach, ANN paradigm was used for modeling the inner relation between the input and the output PLS score vectors. In addition, the capability of the developed method in predicting the biodiesel CN was compared with the basal PLS method. The accuracy of the developed approaches for computing the biodiesel CN was assessed using three statistical criteria, i.e., coefficient of determination (R2), mean-squared error (MSE), and percentage error (PE). The ANN-adapted PLS method predicted the biodiesel CN with an R2 value higher than 0.99 demonstrating the fidelity of the developed model over the classical PLS method with a markedly lower R2 value of about 0.85. In order to facilitate the use of the proposed model, an easy-to-use computer program was also developed on the basis of ANN-adapted PLS method for determining the biodiesel CN from its FAMEs profile.